Driver does not wake GPU properly after suspend (Ubuntu 18.10 with branch 390, 410 and 415)

anon84143002 · November 28, 2018, 9:29am

Issue on Ubuntu bugtracker:

Some added info. This bug is present in:
driver : nvidia-driver-415 - third-party free recommended
driver : nvidia-driver-410 - third-party free
driver : nvidia-driver-390 - distro non-free

GPU info:
== /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0 ==
modalias : pci:v000010DEd00001C8Csv00001028sd0000087Cbc03sc02i00
vendor : NVIDIA Corporation
model : GP107M [GeForce GTX 1050 Ti Mobile]

When in the faulty state nvidia-smi displays a 100% GPU Utilization.
Using prime-select intel to switch GPU and killing X gives back your display environment using onboard GPU.

If needed I’ll run a nvidia-bug-report.sh.

generix · November 28, 2018, 12:55pm

Please try with kernel parameters
acpi_osi=! acpi_osi=“Windows 2009”

pauljohn · November 28, 2018, 4:38pm

The newest Ubuntu 18.10 kernel makes video fail in many ways for me. Before chasing video problems, try the older kernel

https://askubuntu.com/q/1096250/495717

There’s some kind of really important change in the kernels that come after

$ uname -a
Linux delllap-16 4.18.0-10-generic #11-Ubuntu SMP Thu Oct 11 15:13:55 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux

Upgrades after that one cause me to have 1) black screen fail to start display manager and 2) failure to light up after suspend. I gave up trying to fix as soon as I realized the kernel update caused the problems.

I’m not running any off-the-path nvidia drivers. Just xserver-xorg-video-nvidia-390 from Ubuntu 18.10 repositories.

pauljohn · November 28, 2018, 4:41pm

The newest Ubuntu 18.10 kernel makes video fail in many ways for me. Before chasing video problems, try the older kernel

https://askubuntu.com/q/1096250/495717

There’s some kind of really important change in the kernels that come after

$ uname -a
Linux delllap-16 4.18.0-10-generic #11-Ubuntu SMP Thu Oct 11 15:13:55 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux

Upgrades after that one cause me to have 1) black screen fail to start display manager and 2) failure to light up after suspend. I gave up trying to fix as soon as I realized the kernel update caused the problems.

I’m not running any off-the-path nvidia drivers. Just xserver-xorg-video-nvidia-390 from Ubuntu 18.10 repositories.

anon84143002 · November 28, 2018, 9:08pm

With 410 drivers and kernel 4.18.0-11-generic this issue is better solved with kernel parameters:
acpi_rev_override=1 acpi_osi=Linux nouveau.modeset=0 pcie_aspm=force drm.vblankoffdelay=1 scsi_mod.use_blk_mq=1 nouveau.runpm=0 mem_sleep_default=deep
in stead of:
acpi_osi=! acpi_osi="Windows 2009"

This because of the touchpad and other devices also being impacted negatively.

So: Is this a kernel issue or a nvidia driver issue?

generix · November 28, 2018, 9:38pm

It’s a PCI kernel bug:
[url]Invalid Bug ID

pauljohn · November 30, 2018, 4:32am

I just tested this grub config file on Ubuntu with kernel 4.18.0-11-generic and the problem with black screen on resume from suspend is solved, AS LONG AS the display manager is lightdm. gdm3 is still a black screen for me in boot and resume.

GRUB_DEFAULT=0
GRUB_TIMEOUT_STYLE=menu

GRUB_TIMEOUT=0

GRUB_DISTRIBUTOR=lsb_release -i -s 2> /dev/null || echo Debian
GRUB_CMDLINE_LINUX_DEFAULT=“nosplash”
GRUB_CMDLINE_LINUX=“nouveau.blacklist=1 acpi_rev_override=1 acpi_osi=Linux nouveau.modeset=0 pcie_aspm=force drm.vblankoffdelay=1 scsi_mod.use_blk_mq=1 nouveau.runpm=0 mem_sleep_default=deep”

With this (after update-grub) and prime-select nvidia, suspend problem solved. I’ve tested several times today.

fiedler.max · October 6, 2019, 2:18pm

Facing a similar issue. I tried @pauljohn 's workaround but I still face the same issue.

grep on the relevant details of my syslog gives me the following errors (was the same before I updated my grub file according to @pauljohn’s suggestions) (however, I am still using gdm3 as display manager)

Oct  6 15:44:50 nomitri-dl-laptop kernel: [ 3648.254341] nvidia-modeset: WARNING: GPU:0: Lost display notification (0:0x00000000); continuing
.
Oct  6 15:45:07 nomitri-dl-laptop kernel: [ 3665.047242] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57d:0:0
Oct  6 15:45:09 nomitri-dl-laptop kernel: [ 3667.047261] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:1:0
Oct  6 15:45:11 nomitri-dl-laptop kernel: [ 3669.047546] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57d:0:0
Oct  6 15:45:13 nomitri-dl-laptop kernel: [ 3671.047563] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:1:0
Oct  6 15:45:15 nomitri-dl-laptop kernel: [ 3673.047797] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57d:0:0
Oct  6 15:45:17 nomitri-dl-laptop kernel: [ 3675.047810] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:1:0
Oct  6 15:45:30 nomitri-dl-laptop kernel: [ 3688.509371] nvidia-modeset: WARNING: GPU:0: Lost display notification (0:0x00000000); continuing.
Oct  6 15:45:33 nomitri-dl-laptop kernel: [ 3691.512220] nvidia-modeset: WARNING: GPU:0: Lost display notification (0:0x00000000); continuing.
Oct  6 15:48:30 nomitri-dl-laptop kernel: [ 3867.774572] INFO: task nvidia-modeset/:697 blocked for more than 120 seconds.
Oct  6 15:48:30 nomitri-dl-laptop kernel: [ 3867.774575] nvidia-modeset/ D    0   697      2 0x80000000
Oct  6 15:48:30 nomitri-dl-laptop kernel: [ 3867.774608]  nvkms_kthread_q_callback+0x65/0xe0 [nvidia_modeset]
Oct  6 15:48:30 nomitri-dl-laptop kernel: [ 3867.774612]  _main_loop+0x76/0x140 [nvidia_modeset]
Oct  6 15:48:30 nomitri-dl-laptop kernel: [ 3867.774617]  ? _raw_q_schedule+0x80/0x80 [nvidia_modeset]
Oct  6 15:50:30 nomitri-dl-laptop kernel: [ 3988.606189] INFO: task nvidia-modeset/:697 blocked for more than 120 seconds.
Oct  6 15:50:30 nomitri-dl-laptop kernel: [ 3988.606192] nvidia-modeset/ D    0   697      2 0x80000000
Oct  6 15:50:30 nomitri-dl-laptop kernel: [ 3988.606213]  nvkms_kthread_q_callback+0x65/0xe0 [nvidia_modeset]
Oct  6 15:50:30 nomitri-dl-laptop kernel: [ 3988.606217]  _main_loop+0x76/0x140 [nvidia_modeset]
Oct  6 15:50:30 nomitri-dl-laptop kernel: [ 3988.606224]  ? _raw_q_schedule+0x80/0x80 [nvidia_modeset]

Any idea on how to fix this would be highly appreciated. This bug is getting really annoying.

I am running Ubuntu 18.04 on a Lenovo Legion Y740 with an NVIDIA RTX 2070 with CUDA 10.0 installed

$ uname -a
Linux nomitri-dl-laptop 5.0.0-31-generic #33~18.04.1-Ubuntu SMP Tue Oct 1 10:20:39 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux

nvidia-smi
Sun Oct 6 16:18:16 2019
±----------------------------------------------------------------------------+
| NVIDIA-SMI 430.26 Driver Version: 430.26 CUDA Version: 10.2 |

generix · October 6, 2019, 2:37pm

Please check if enabling csm in bios works around it.

fiedler.max · October 6, 2019, 2:47pm

@generix thanks for the quick response.
how would that help?
I will try it, but I am always eager to learn on the reasoning behind things. Could you please, also explain me a bit more about the nature of the error message?
That way, I might be able to help myself in the future.

generix · October 6, 2019, 2:50pm

IDK, it’s just an observation that on some systems running into

ERROR: GPU:0: Idling display engine timed out

enabling (not using) the csm worked around it. Maybe due to additional vga bios initialization.

fiedler.max · October 6, 2019, 3:11pm

I see, the problem is that on my Lenovo Legion, switching to legacy is not so trivial

https://forums.lenovo.com/t5/ThinkPad-T400-T500-and-newer-T/Unable-to-select-UEFI-Legacy-Boot-in-BIOS-ThinkPad-T480/m-p/43612T88

As the necessary step

In BIOS go to Config > Storage > Controller Mode > change to AHCI mode from RST mode

Now you can able to change secure boot setting and able to enable Legacy Mode.

would clear all my data, this is not a solution to try out on a hunch for me.

Do you have any other ideas on what I could try or what the error and the warning before the error nvidia-modeset: WARNING: GPU:0: Lost display notification (0:0x00000000); continuing
might hint at as the underlying cause of this behavior?

adityagupte95 · November 5, 2019, 7:12pm

I am facing the same issue as @fiedler.max. I am using the same device and config as @fiedler.max and every time face same error of “idling engine timed out while resuming ubuntu 18.04”. The only way is to reboot the device.
nvidia-bug-report.log.gz (1.96 MB)

generix · November 5, 2019, 10:19pm

Please check for a bios update.

adityagupte95 · November 23, 2019, 7:24pm

I did try to update bios, it tells me i am on the latest version available for my laptop, anything else that i can try?

niyoceles3 · July 20, 2020, 8:04pm

Thank you pauljohn, you saved my time.

silocoder · December 11, 2020, 9:56pm

I have same issue does not wake up after suspend and needs a reboot. I’ve tried @pauljohn solution but my display manager is sddm.

this is my grub

GRUB_DEFAULT="saved"
GRUB_TIMEOUT_STYLE="hidden"
GRUB_TIMEOUT="20"
GRUB_DISTRIBUTOR="`lsb_release -i -s 2> /dev/null || echo Debian`"
GRUB_CMDLINE_LINUX_DEFAULT="nosplash"
GRUB_CMDLINE_LINUX="nouveau.blacklist=1 acpi_rev_override=1 acpi_osi=Linux nouveau.modeset=0 pcie_aspm=force drm.vblankoffdelay=1 scsi_mod.use_blk_mq=1 nouveau.runpm=0 mem_sleep_default=deep"

originally GRUB_CMDLINE_LINUX was blank.

Any other ideas would be appreciated.

I am running ubuntu 20.04, Plasma KDE 5

Graphics:  Device-1: Intel 4th Gen Core Processor Integrated Graphics driver: i915 v: kernel 
           Device-2: NVIDIA GK104M [GeForce GTX 870M] driver: nvidia v: 455.45.01 
           Display: x11 server: X.Org 1.20.8 driver: modesetting,nvidia unloaded: fbdev,nouveau,vesa 
           resolution: 2880x1620~60Hz 
           OpenGL: renderer: GeForce GTX 870M/PCIe/SSE2 v: 4.6.0 NVIDIA 455.45.01

I

filipprayden · December 26, 2020, 1:11pm

The same is on Ubuntu 20.04, AMD Ryzen 7 4800H & Nvidia 1660 ti Max-Q

vlad1777d · February 22, 2021, 12:11pm

Same thing with HP Pavilion Power cb015.
Linux Mint 20.1 Cinnamon (Ubuntu 20.04 LTS).

Tried Kernels: 5.4.0-65, 5.8.0.43.
No options helped to avoid this issue with 460 driver.
But problem is missing with 450 and 390 drivers.

Seems to be some kind of regression =)
I’m not surprised, because always have problems with different nvidia cards and different laptops.
Good thing - that they’re not critical, main workflow isn’t corrupted.

Btw, everything works fine on Fedora 33 Cinnamon with 460 driver and 5.10.16 kernel…